Dear all, Beside talking to several other speakers or attendees interested in Mobile and DeviceMap last week, the DeviceMap session had a small but exclusive audience (including representatives or speakers from the 2 top ApacheCon sponsors)
Also nice to meet Bertrand again in person though he had many responsibilities and "hats" so he did not manage to see the DeviceMap session. The Q&A on what we should focus on brought some constructive input - Using JSON may be a bit premature, especially due to a lack of standardized JSON Schema similar to XSL. http://json-schema.org/ may work on it, but there is no standardization like JSON itself by ECMA. - Trying to improve device recognition by adding new or updated information (e.g. an OS upgrade or new/different browser) based on e.g. a "patch" from the User Agent is a good idea. If someone upgraded their OS to Android 5 or Windows 10 they don't want to still see the OS their device was once sold with - Where technical and administrative means at Apache allow it, we should try to crowd-source gathering new device signatures. How that's best done (probably on the same VM instances that run web-based example apps now) we shall figure out, but other meetings in Budapest like Apache BarCamp also came to the conclusion, that using such service for a particular project should not require signing a full Apache Contributor Agreement first. It's the first level of contribution like filing a JIRA ticket or proposing a patch by email. At most the web app should contain a simple click-through agreement or login where users confirm they send data that belongs to them and not copied something that may be protected or belong to others. A minor detail was, that the term "client" seems to mislead a few who think that part of software runs on the actual device (like Cordova) at least that was another question I answered during my talk. Major commercial competitors like DeviceAtlas also do call some of these components "Client", but in many cases the term "API" seems to dominate over "Client". We may want to adjust that in some cases, if not in actual classes, then mabe when releasing new artifacts. I plan to release the VB.NET "Client" (probably won't change the name there yet) similar to C# in the course of this week or early next week. Most attendees from companies like Red Hat were naturally strong users of Java, so they were not so interested in the .NET part though I did a brief demo of it, too. If you have immediate feedback, please try to CC or BCC. I'll keep an eye on the archive every now and then, so if it's not extremely urgent, just reply to the list. Regards, Werner
